AgentQuartz: Real-Time Claude and Cursor Usage Monitoring from Your macOS Menu Bar

A Usage Monitoring Assistant Built for AI Programming Tools

As AI programming tools like Claude and Cursor become deeply integrated into developers' daily workflows, a new pain point has gradually emerged: how do you keep track of your usage quota at all times? Constantly switching to a browser to check usage limits not only breaks your flow state but also significantly undermines what should be an efficient development experience.

To understand the severity of this pain point, you need to understand the quota mechanisms of these AI tools. Claude Pro users have a monthly conversation cap, and when usage approaches the limit, they get downgraded to slower models or temporarily lose access. Cursor Pro sets monthly fast request quotas for its AI code completion and Chat features (e.g., 500 fast requests per month), after which users must wait or switch to slow mode. These limits directly impact developers' work rhythm — if your quota runs out without you realizing it, you might suddenly lose AI assistance at a critical coding moment, causing serious work disruption.

AgentQuartz is a macOS menu bar app targeting precisely this niche need. Built on a "Local-first" core philosophy, it lets users check Claude and Cursor usage in real time with just a glance at the menu bar — no browser tab required. The product received 107 upvotes and 14 comments on Product Hunt, ranking #14 for the day, demonstrating genuine demand from the developer community for this type of efficiency tool.

Local-First Architecture: Dual Guarantee of Privacy and Lightweight Design

No Cloud Account, All Data Stays on Your Machine

AgentQuartz's most noteworthy design decision is its privacy architecture. Unlike many SaaS tools that require cloud account registration, AgentQuartz explicitly states "no AgentQuartz cloud account" — all settings are saved on the user's own machine.

It's worth explaining the deeper meaning of the "Local-first" design philosophy here. This concept was systematically articulated by the Ink & Switch lab in their 2019 research paper. Its core principle is: the primary copy of data is stored on the user's local device, the app can function normally without a network connection, and the cloud (if present at all) serves only as an auxiliary role for sync and backup. Unlike the traditional SaaS model where data is entirely hosted on vendor servers, Local-first apps give users complete ownership and control over their data. In AgentQuartz's case, this means sensitive information like API keys, usage frequency, and quota preferences never leaves the local machine.

This design yields two direct benefits: First, privacy and security — user data never gets uploaded to third-party servers, which is especially important for developers who prioritize data sovereignty. Second, response speed — localized data processing means near-zero latency for usage queries. In an era where AI tools overwhelmingly rely on cloud synchronization, this "counter-trend" local-first strategy actually creates differentiated competitiveness.

Restrained Design in Menu Bar Form Factor

As a menu bar app, AgentQuartz follows the design philosophy of a classic category of tools in the macOS ecosystem: don't compete for screen space, quietly reside in the system tray, and be available at a glance when needed. This "invisible presence" product form perfectly aligns with developers' core desire to minimize context switching and maintain focus.

The macOS menu bar app is a unique and mature product form in the Apple ecosystem, with classic examples including 1Password's quick access, iStat Menus' system monitoring, and Bartender's menu bar management. The design philosophy of these apps stems from the "Ambient Awareness" concept — information should be like a thermometer, available at a glance when needed and completely non-intrusive when not. Technically, macOS provides developers with the ability to reside in the menu bar through the NSStatusItem API. Apps typically occupy only a tiny icon position, expanding into a dropdown panel with detailed information when clicked. AgentQuartz's choice of this form factor shows the developer deeply understands that the optimal interaction pattern for "status monitoring needs" isn't opening a full window, but integrating into the OS's ambient layer.

Pricing Strategy: Free Core, One-Time Pro Unlock

AgentQuartz employs a remarkably user-friendly business strategy. Core features are "Free forever for essentials," ensuring basic usage monitoring is available to all users. The Pro version requires only a one-time payment of $3 to unlock Tray customization features.

This "one-time purchase" rather than subscription-based pricing feels refreshingly sincere in today's increasingly subscription-dominated software market. In recent years, as subscription fatigue has spread among consumers — according to Paddle's 2023 report, approximately 35% of developer tool users explicitly prefer one-time payment models — the buy-once model has experienced a notable revival in indie developer circles. This strategy is particularly suitable for small tools with clear functionality and low maintenance costs. The $3 price point falls in the "impulse purchase" range, where users barely need to go through complex value assessment before paying. Similar successful examples include Raycast's free+Pro model and CleanShot X's one-time purchase strategy.

For heavy AI tool users who monitor their usage quota daily, this investment is practically negligible.

Important Note: An Unofficial Third-Party Monitoring Tool

You might not have noticed, but AgentQuartz explicitly labels itself as "Unofficial" in its description. This means it's not a product released by Anthropic (the company behind Claude) or Cursor's team, but rather a third-party tool built by independent developer Marco Calderon.

This attribute brings a dual nature. On one hand, third-party tools are typically more flexible and iterate faster, able to quickly respond to users' niche needs. On the other hand, they face dependency risks on official APIs or data interfaces — if the official platforms adjust their usage query mechanisms, the tool's availability could be affected.

From a technical perspective, third-party monitoring tools typically obtain usage data through several approaches: calling officially provided public APIs, parsing user account page data (i.e., web scraping), or simulating login via browser cookies to retrieve information. Each approach faces different stability risks. Anthropic currently provides a usage endpoint through its API, but the interface's stability and backward compatibility have no formal SLA guarantee. If the tool relies on page parsing to obtain data, any frontend redesign could cause functionality to break. Historically, events like Twitter (now X) and Reddit drastically tightening third-party API access have repeatedly proven these risks are not merely theoretical concerns.

For potential users, understanding this helps establish reasonable usage expectations.

An Emerging Trend in the AI Programming Tool Ecosystem

Although AgentQuartz is a compact efficiency tool, it reflects an important trend in the AI programming era: a rich third-party tool ecosystem is forming around mainstream AI tools like Claude and Cursor.

This ecosystem's breadth extends far beyond the single scenario of usage monitoring. The current third-party ecosystem around AI programming tools already includes: prompt management tools (like PromptLayer, Langfuse) helping developers track and optimize AI interaction history; cost analysis tools (like OpenRouter's cost tracking features) helping teams control AI spending; workflow automation tools (like n8n, Make integrations with AI APIs) embedding AI capabilities into more complex development processes; and various IDE plugins continuously expanding AI assistants' capabilities.

From usage monitoring to workflow enhancement, indie developers are keenly capturing various subtle pain points in AI tool usage and solving them with lightweight products. This "tools for tools" phenomenon is precisely a sign that a technology ecosystem is maturing. There are precedents in software history — similar phenomena appeared in the early AWS ecosystem (such as Datadog's rise as a third-party alternative to CloudWatch) and the Git ecosystem (various Git GUI clients), typically signaling that the underlying platform has reached sufficient user scale and usage depth to support the commercialization of upper-layer tools.

As more and more developers rely on AI-assisted programming, how to better manage, monitor, and optimize the use of these tools itself becomes a market worth cultivating deeply.

For developers who work with Claude and Cursor every day, AgentQuartz provides a low-cost, privacy-focused, high-efficiency solution. It may not be a "disruptive" product, but it's precisely these small tools that precisely solve real problems that form an important part of developers' daily experience.

Key Takeaways